AJCN. First published ahead of print May 21, 2014 as doi: 10.3945/ajcn.113.076109.

Nut consumption and risk of type 2 diabetes, cardiovascular disease, and all-cause mortality: a systematic review and meta-analysis1–4 Cheng Luo, Yan Zhang, Yusong Ding, Zhilei Shan, Sijing Chen, Miao Yu, Frank B Hu, and Liegang Liu ABSTRACT Background: Epidemiologic studies have shown inverse associations between nut consumption and diabetes, cardiovascular disease (CVD), and all-cause mortality, but results have not been consistent. Objective: We assessed the relation between nut intake and incidence of type 2 diabetes, CVD, and all-cause mortality. Design: We searched PubMed and EMBASE for all prospective cohort studies published up to March 2013 with RRs and 95% CIs for outcomes of interest. A random-effects model was used to pool risk estimates across studies. Results: In 31 reports from 18 prospective studies, there were 12,655 type 2 diabetes, 8862 CVD, 6623 ischemic heart disease (IHD), 6487 stroke, and 48,818 mortality cases. The RR for each incremental serving per day of nut intake was 0.80 (95% CI: 0.69, 0.94) for type 2 diabetes without adjustment for body mass index (BMI); with adjustment, the association was attenuated [RR: 1.03; 95% CI: 0.91, 1.16; NS]. In the multivariable-adjusted model, pooled RRs (95% CIs) for each serving per day of nut consumption were 0.72 (0.64, 0.81) for IHD. 0.71 (0.59, 0.85) for CVD, and 0.83 (0.76, 0.91) for all-cause mortality. Pooled RRs (95% CIs) for the comparison of extreme quantiles of nut intake were 1.00 (0.84, 1.19; NS) for type 2 diabetes, 0.66 (0.55, 0.78) for IHD, 0.70 (0.60, 0.81) for CVD, 0.91 (0.81, 1.02; NS) for stroke, and 0.85 (0.79, 0.91) for all-cause mortality. Conclusions: Our meta-analysis indicates that nut intake is inversely associated with IHD, overall CVD, and all-cause mortality but not significantly associated with diabetes and stroke. The inverse association between the consumption of nuts and diabetes was attenuated after adjustment for body mass index. These findings support recommendations to include nuts as part of a healthy dietary pattern for the prevention of chronic diseases. Am J Clin Nutr doi: 10.3945/ajcn.113.076109.

INTRODUCTION

Cardiovascular disease (CVD)5 is a major public health concern in both developed and developing countries. The American College of Cardiology predicts that the number of people diagnosed with CVD will double to 25 million in the United States alone by 2050 (1). In individuals with CVD, stroke is associated with permanent disability, mortality, and major direct and indirect costs because of functional impairments (2, 3). The increase in diabetes over the past 20 y has been dramatic; the current prevalence is now estimated to be nearly 6.4% worldwide (4). For these reasons, the primary prevention of these chronic diseases is imperative.

Dietary factors are one of the major determinants of both type 2 diabetes and CVD. Of various dietary factors, nuts have received increasing attention because they are a good source of macronutrients and micronutrients. Nuts are rich in unsaturated fatty acids, fiber, high-quality vegetable protein, and minerals (eg, magnesium and potassium). Nuts also have high contents of bioactive compounds, such as polyphenols, tocopherols, phytosterols, and phenolics (5–7). Walnuts are especially rich in polyunsaturated acids, including a-linolenic acid. Previous clinical trials and epidemiologic studies have shown that nuts have beneficial effects on various mediators of chronic diseases, including lipid concentrations (8), inflammation (9), insulin resistance (10), and blood pressure (BP) (11). A previous review concluded that the consumption of nuts $5 times/wk was associated with a reduced incidence of coronary heart disease (IHD) (12). In the past several years, additional studies have been published on CVD, diabetes, and all-cause mortality. However, the epidemiologic evidence has not been consistent. Therefore, we conducted this meta-analysis on all published cohort studies to date to quantify the relation between nut consumption and risk of type 2 diabetes, CVD, and all-cause mortality. METHODS

Data sources and searches We searched PubMed (http://www.ncbi.nlm.nih.gov/pubmed/) and EMBASE (http://www.embase.com/) databases for all English-language 1 From the Department of Nutrition and Food Hygiene, Hubei Key Laboratory of Food Nutrition and Safety, Tongji Medical College, Hua Zhong University of Science and Technology, Wuhan, China (CL, YZ, YD, ZS, SC, MY, and LL); the Ministry of Education Key Laboratory of Environment and Health, School of Public Health, Tongji Medical College, Wuhan, China (CL, YZ, YD, ZS, SC, MY, and LL); and the Department of Nutrition and Epidemiology, Harvard School of Public Health, Boston, MA (FBH). 2 Funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. 3 Supported by the National Science and Technology Support Program (2012BAI02B02) of China and the National Natural Science Foundation of China (NSFC 81072291). 4 Address reprint requests and correspondence to L Liu, Tongji Medical College, 13 Hangkong Road, Wuhan, China. E-mail: [email protected]; or FB Hu, Harvard School of Public Health, 665 Huntington Avenue, Boston, MA 02115. E-mail: [email protected]. 5 Abbreviations used: BP, blood pressure; CVD, cardiovascular disease; IHD, ischemic heart disease; MI, myocardial infarction. Received September 20, 2013. Accepted for publication March 17, 2014. doi: 10.3945/ajcn.113.076109.

Am J Clin Nutr doi: 10.3945/ajcn.113.076109. Printed in USA. Ó 2014 American Society for Nutrition

Copyright (C) 2014 by the American Society for Nutrition

1 of 14

2 of 14

LUO ET AL

prospective cohort studies published in peer-reviewed journals through March 2013. Search terms included human and nuts or almond or cashew or peanut or pecan or pine nut or pistachio nut or macadamia nut or hazelnut or walnut and diabetes or CVD or myocardial infarction (MI) or coronary or stroke or death or mortality or mortalities or fatal and risk or Cox or hazard or odds. We also reviewed references from retrieved articles to find additional studies.

Study selection We selected studies on the basis of the following criteria: 1) a prospective cohort study, 2) nut-consumption exposure, 3) the outcome of interest (ie, type 2 diabetes, CVD, or all-cause mortality), and 4) RRs with 95% CIs. On the basis of reviews of titles and abstracts, we identified 122 pertinent articles (Figure 1). We excluded reviews, case-control studies, editorials, nonhuman studies, dietary pattern reports, and articles that examined associations with foods other than nuts. If data were published more than once, we included the largest number of incident cases from the study. We included the following 18 studies in the meta-analysis: 5 studies (13–17) stratified analyses by sex; one study (18) used

2 independent cohorts; one study (19) reported findings for IHD, CVD, and mortality; 3 studies (20–22) had 2 different outcomes; one study (23) had results for hemorrhagic and ischemic strokes; 2 studies (24, 25) had outcomes for fatal IHD and nonfatal MI; and 5 studies (26–30) had one outcome. One study stratified diabetes by BMI (in kg/m2) (31). We pooled data on fatal IHD and nonfatal MI to get an overall estimate. In total, 31 reports from 18 studies were included in the meta-analysis as follows: 5 reports were on diabetes, 6 reports were on IHD, 5 reported were on stroke, 4 reports were on total CVD, and 11 reports were on all-cause mortality. Data extraction and quality assessment Two independent investigators extracted data, including the name of the first author, year of publication, study title, sample size, sex and age range, type of nut (eg, with and without peanut butter), length of follow-up, country of study, inclusion and exclusion criteria, specific outcomes, number of cases, types of nuts consumed (ie, walnuts and total tree nuts), covariates that were adjusted, and RRs with corresponding 95% CIs for all categories of nut consumption. We extracted fully adjusted estimates. In addition, we extracted multivariable-adjusted estimates without adjustment for BMI when available. We used the Meta-analysis of Observational Studies in Epidemiology (32) for quality assessment. This scoring system awards a maximum of 5 points as follows: study eligibility (1 point if inclusion and exclusion criteria are included), exposure (1 point if dietary assessment is validated and 1 point if nut intake is appropriately categorized), outcome (1 point for the assessment of outcome according to accepted clinical criteria), and statistical analysis (1 point for the control of confounding factors). Statistical analysis

FIGURE 1. Flowchart of study selection. PubMed, http://www.ncbi.nlm. nih.gov/pubmed/; EMBASE, http://www.embase.com/. CVD, cardiovascular disease; IHD, ischemic heart disease.

Results stratified by sex and stroke subgroups were treated as 2 separate reports. A random-effects model was used to calculate summary RRs and 95% CIs for a 1-serving/d increase of nut intake. For each study that used different cutoffs for nut categories, we evaluated the dose-response relation by computing RRs with 95% CIs for a 1-serving/d increment of nut consumption. For studies that provided results in servings per day, we used the original data to estimate RRs with 95% CIs for the 1-serving/d increase in nut intake. If outcomes were only shown by categories of nut intake, we used the method proposed by Greenland and Longnecker (33) and Orsini et al (34) to estimate RRs with 95% CIs for the 1-serving/d increase in nut intake. When the median or mean consumption per category was not reported, we assigned the midpoint of upper and lower boundaries in each category as the median consumption. For the highest consumption category, we assumed that the lower boundary plus a 25% increment was the median intake. For studies that reported categorical data in grams, we first converted nut intake into servings per day with a standard conversion (1 serving = 28 g) (35). We also used a random-effects model to calculate summary RRs and 95% CIs for highest compared with lowest intakes of nut consumption. We used the I2 statistic to assess heterogeneity. Low, moderate, and high degrees corresponded to I2 values of 25%, 50%, and 75% (36, 37), respectively. The Egger test was used to assess publication bias (38). We performed subgroup analyses on

F

F

M

F

F

Villegas (27), 2008, The Shanghai Women’s Health Study, China

Kochar (28), 2010, PHS, United States

Pan (18), 2013, NHS, United States

Pan (18), 2013, NHS II, United Staetes

Sex

Parker (26), 2003, IWHS, United States

First author (ref), year of publication, study name, location

20–45

30–55

41–87

18

22

21.1

4.6

y 11

y 55–69

40–70

Follow-up time

Age

Diabetes 4098/159,9667

Diabetes 5121/116,4248

Diabetes 1828/20,224

Diabetes 1608/64,227

Diabetes 1831/134,486

No. of cases/subjects ,1/mo; 1.00 (1.00, 1.00) ,1/wk; 0.98 (0.87, 1.10) 1–4/wk; 1.06 (0.93, 1.22) $5/wk; 1.51 (1.13, 2.04) 0.1 g/d; 1.00 (1.00, 1.00) 0.4 g/d; 0.80 (0.69, 0.94) 0.7 g/d; 0.95 (0.82, 1.11) 1.4 g/d; 0.79 (0.68, 0.92) 3.1 g/d; 0.80 (0.68, 0.93) 0/wk; 1.00 (1.00, 1.00) ,1/wk; 1.06 (0.93, 1.20) 1/wk; 1.10 (0.95, 1.26) 2–4/wk; 0.97 (0.82, 1.14) 5–6/wk; 0.99 (0.76, 1.30) .7/wk; 0.87 (0.61, 1.24) Fully adjusted model without BMI 0/wk; 1.00 (1.00, 1.00) ,1/wk; 0.97 (0.91, 1.04) 1/wk; 1.00 (0.91, 1.09) 2–4/wk; 0.91 (0.80, 1.02) $5/wk; 0.87 (0.77, 0.99) Fully adjusted model 0/wk; 1.00 (1.00, 1.00), ,1/wk; 0.99 (0.92, 1.05) 1/wk; 1.05 (0.96, 1.15) 2–4/wk; 0.98 (0.87, 1.11) $5/wk; 1.00 (0.87, 1.14) Fully adjusted model without BMI 0/wk; 1.00 (1.00, 1.00) ,1/wk; 0.95 (0.88, 1.02) 1/wk; 0.89 (0.79, 0.99) 1–4/wk; 0.86 (0.73, 1.00) $5/wk; 0.77 (0.64, 0.92) Fully adjusted model 0/wk; 1.00 (1.00, 1.00) ,1/wk; 0.99 (0.92, 1.06) 1/wk; 1.00 (0.89, 1.12) 1–4/wk; 1.00 (0.86, 1.17) $5/wk; 1.02 (0.85, 1.23)

Dietary category and RR (95% CI)

Same as above

Validated FFQ 130 items/symptoms plus FPG, OGTT, or use of antidiabetic medication

Validated FFQ 19 items/self-report

Validated FFQ 77 items/symptoms plus FPG, OGTT, or taking antidiabetic medication

Validated FFQ 127 items/self-report

Exposure and case ascertainment2

TABLE 1 Characteristics of 18 prospective cohort studies on nut consumption and diabetes, IHD, stroke, CVD, and all-cause mortality events1

Same as above

(Continued)

Age, smoking, family history of diabetes and physical activity, total energy intake, alcohol intake, BMI, diet, and magnesium

Age, smoking, randomization arm, consumption of breakfast, cereal, dairy, and red meat, physical activity, BMI, and history of hypertension

Age, energy intake, BMI, waist-to-hip ratio, smoking, alcohol, vegetable intake, fiber, physical activity, income, education, occupation, and hypertension

Age, smoking, family history of diabetes and physical activity, total energy intake, alcohol intake, BMI, diet, et al

Covariates in fully adjusted model

META-ANALYSIS: NUTS AND DIABETES, CVD, AND MORTALITY

3 of 14

Both

M

F

F

Albert (25), 2002, PHS, United States

Blomhoff (19), 2006, IWHS, United States

Li (21), 2009, NHS, United States

Sex

Fraser (24), 1992, AHS, United States

First author (ref), year of publication, study name, location

TABLE 1 (Continued )

15

22

55–69

52.8 6 8.5

17

6

.25

40–84

Follow-up time

Age

CVD death 635 MI 452/6309

IHD death 948 CVD death 1675 Total death 5451/31,778

Nonfatal MI 1037 Fatal IHD 566/21,454

Nonfatal MI 134 Fatal IHD 260/31,208

No. of cases/subjects Fatal IHD ,1/wk; 1.00 (1.00, 1.00) 1–4/wk; 0.74 (0.49, 1.11) .5/wk; 0.52 (0.30, 0.87) Nonfatal MI ,1/wk; 1.00 (1.00, 1.00) 1–4/wk; 0.77 (0.61, 0.98) .5/wk; 0.67 (0.51, 0.88) Fatal IHD ,1/mo; 1.00 (1.00, 1.00) 1/wk; 0.90 (0.67, 1.22) 1–3/mo; 0.89 (0.67, 1.16) $2/wk; 0.70 (0.50, 0.98) Nonfatal MI ,1/mo; 1.00 (1.00, 1.00) 1–3/mo; 1.22 (1.00, 1.51) 1/wk; 1.20 (0.96, 1.50) $2/wk; 1.04 (0.82, 1.33) IHD death 0/wk; 1.00 (1.00, 1.00) ,1/wk; 1.03 (0.84, 1.26) 1–4/wk; 0.82 (0.68, 0.98) $5/wk; 0.71 (0.55, 0.91) CVD death 0/wk; 1.00 (1.00, 1.00) ,1/wk; 1.00 (0.86, 1.17) 1–4/wk; 0.84 (0.73, 0.96) $5/wk; 0.72 (0.60, 0.88) Total death 0/wk; 1.00 (1.00, 1.00) ,1/wk; 0.93 (0.85, 1.02) 1–4/wk; 0.88 (0.81, 0.95) $5/wk; 0.89 (0.81, 0.99) Nonfatal MI 0/wk; 1.00 (1.00, 1.00) ,1/wk; 0.63 (0.41, 0.96) 2–4/wk; 0.74 (0.49, 1.13) $5/wk; 0.56 (0.33, 0.97) CVD 0/wk; 1.00 (1.00, 1.00) ,1/wk; 0.72 (0.50, 1.02) 2–4/wk; 0.80 (0.56, 1.14) $5/wk; 0.56 (0.36, 0.89)

Dietary category and RR (95% CI)

Peanut butter + nuts, semiquantitative FFQ 131 items/ADA diagnostic criteria

Peanut butter + nuts FFQ 127 items/ NS

FFQ 20 items/self-report and medical records

Semiquantitative FFQ 65 items/ medical records, autopsy report, death certificate

Exposure and case ascertainment2

(Continued)

Age, time period, total energy, cereal fiber, alcohol, trans fat, BMI, smoking, menopausal status, parental history of early myocardial infarction, multivitamin, vitamin E, aspirin use, and exercise

BMI, waist:hip ratio, education, physical activity, use of estrogens, multivitamin supplements, alcohol, whole grain, refined grain, red meat, fish, seafood, and total fruit and vegetable intake

Age, aspirin and carotene treatment, evidence of CVD, BMI, smoking, history of diabetes, hypercholesterolemia, hypertension, alcohol, exercise, and fish, red meat, fruit, vegetable, and dairy intake

Age, sex, smoking, exercise, relative weight, high blood pressure, and consumption of other foods

Covariates in fully adjusted model

4 of 14 LUO ET AL

F

M

Both

F

Both

Djousse´ (23), 2010, PHS, United States

Bernstein (15), 2012, HPFS, United States

Yaemsiri (30), 2012, WHI-OS, United States

Fraser (14), 1997, AHS, United States

Sex

Bernstein (29), 2010, NHS, United States

First author (ref), year of publication, study name, location

TABLE 1 (Continued )

4



.25

men 26 women 22

21.1

26

Follow-up time

50–79

Men: 40–75 Women: 30–55

41–87

30–55

Age

All-cause mortality Men 257 Women 312/NS

Ischemic stroke 1049/87,025

Total stroke Men 1397/43,150 Women 2633/84,010

Total stroke 1408/21,078

IHD 3162/34,492

No. of cases/subjects

Men ,1/wk; 1.00 (1.00, 1.00) 1–4/ wk; 0.7 (0.4, 1.4)

Hemorrhagic stroke 0/wk; 1.00 (1.00, 1.00) ,1/wk; 1.13 (0.78, 1.62) 1/wk; 1.05 (0.70, 1.58) 2–4/wk; 0.49 (0.27, 0.89) 5–6/wk; 1.50 (0.79, 2.84) $7/wk; 1.84 (0.95, 3.57) Ischemic stroke 0/wk; 1.00 (1.00, 1.00) ,1/wk; 0.93 (0.65, 1.34) 1/wk; 0.86 (0.74, 1.01) 2–4/wk; 0.94 (0.79, 1.11) 5–6/wk; 0.97 (0.80, 1.18) $7/wk; 1.06 (0.79, 1.43) Men 0.0/wk; 1.00 (1.00, 1.00) 0.5/wk; 0.94 (0.79, 1.12) 1.0/wk; 0.95 (0.80, 1.13) 1.8/wk; 1.01 (0.86, 1.20) 4.2/wk; 0.92 (0.77, 1.09) Women 0.0/wk; 1.00 (1.00, 1.00) 0.3/wk; 0.94 (0.83, 1.06) 0.5/wk; 0.91 (0.80, 1.04) 0.8/wk; 0.97 (0.85, 1.10) 2.4/wk; 0.86 (0.75, 0.98) Per 7/wk; 0.89 (0.66, 1.20)

0.00/wk; 1.00 (1.00, 1.00) 0.28/wk; 0.73 (0.65, 0.82) 0.49/wk; 0.91 (0.82, 1.00) 0.84/wk; 0.76 (0.67, 0.84) 2.8/wk; 0.68 (0.60, 0.76)

Dietary category and RR (95% CI)

Semiquantitative FFQ 65 items/NS

FFQ 122 items/by centrally physicians and trained neurologists

Validated FFQ 131 items/medical records

Semiquantitative FFQ 19 items/death certificates and medical records

FFQ 131 items/medical records, symptoms plus electrocardiogram or enzymes [, National Death Index

Exposure and case ascertainment2

(Continued)

Age, race, education, income, smoking, alcohol, history of IHD, AF, aspirin use, diabetes, BMI, BP, vitamin E, fruit, vegetable, fiber intake, and use of cholesterol-lowering, antihypertensive medication The food of interest as well as age, smoking, and exercise

Age, time period, total energy, cereal fiber, alcohol, trans fat, BMI, smoking, menopausal status, parental history of early myocardial infarction, exercise, multivitamin, vitamin E supplement, aspirin use, fruit, vegetables, and other protein sources

Age, BMI, activity, alcohol, family history of MI, hormone use, menopausal status, smoking, aspirin intake, duration of diabetes years, hypertension, hypercholesterolemia, energy intake, cereal fiber, glycemic load, saturated fat, and trans fat Age, aspirin assignment, BMI, alcohol consumption, smoking, fruit and vegetable intake, exercise, breakfast cereal, red meat, fish, dairy consumption, prevalent hypertension, diabetes, atrial fibrillation, and IHD

Covariates in fully adjusted model

META-ANALYSIS: NUTS AND DIABETES, CVD, AND MORTALITY

5 of 14

Both

Both

Both

Both

Mann (20), 1997, Oxford Vegetarian study, United Kingdom

Van den Brandt (17), 2011, the Dutch study, Netherlands

Bao (13), 2013, NHS and HPFS, United States

Sex

Fraser (16), 1997, AHS, United States

First author (ref), year of publication, study name, location

TABLE 1 (Continued )

NHS 30–55 HPFS 40–75

55–69

NHS 30 HPFS 24

24

13.3

12

.85

16–79

Follow-up time

Age

All-cause mortality NHS 16,200 HPFS 11,229 CVD death 6471/3,038,853

All-cause mortality Men 8019 Women 5248/NS

IHD death 64 All-cause mortality 392/10,802

All-cause mortality 1387/NS

No. of cases/subjects 5/wk; 0.6 (0.2, 1.3) Women ,1/wk; 1.00 (1.00, 1.00) 1–4/wk; 0.4 (0.2, 0.9) $5/wk; 0.5 (0.2, 1.2) Men ,1/wk; 1.00 (1.00, 1.00) 1–4/wk; 0.84 (0.64, 1.09) $5/wk; 0.77(0.58, 1.02) Women ,1/wk; 1.00 (1.00, 1.00) 1–4/wk; 0.89 (0.75, 1.06) $5/wk; 0.84 (0.70, 1.01) IHD death ,1/wk; 1.00 (1.00, 1.00) 1–4/wk; 1.19 (0.68, 2.10) $5/wk; 0.87 (0.45, 1.68) All-cause mortality ,1/wk; 1.00 (1.00, 1.00) 1–4/wk; 0.99 (0.79, 1.25) $5/wk; 0.77 (0.58, 1.01) Men 0.0g/d 1.00 (1.00, 1.00) 11.1g/d 0.92 (0.87, 0.98) Women 0.0g/d 1.00 (1.00, 1.00) 6.2g/d 0.95 (0.90, 1.00) Women All-cause mortality 0/wk; 1.00 (1.00, 1.00) ,1/wk; 0.94 (0.90, 0.98) 1/wk; 0.88 (0.83, 0.92) 2–4/wk; 0.85 (0.80, 0.90) 5–6/wk; 0.88 (0.78, 0.98) $7/wk; 0.79 (0.68, 0.91) Men: all-cause mortality 0/wk; 1.00 (1.00, 1.00) ,1/wk; 0.85, 0.96) 1/wk; (0.86, 0.97) 2–4/wk; (0.83, 0.94) 5–6/wk; (0.76, 0.91) $7/wk; 0.80 (0.73, 0.88) CVD death

Dietary category and RR (95% CI)

Validated FFQ 131 items/National Death Index, reports from family members and postal authorities

Semiquantitative FFQ 150 items/by linkage to the Dutch Central Bureau of Genealogy

Semiquantitative FFQ/collected with office for National Statistics and coded blinded

Semiquantitative FFQ/by matching to California State death tapes and the National Death Index

Exposure and case ascertainment2

(Continued)

Age, race, BMI, physical activity, smoking, current multivitamin use, aspirin use, family history of diabetes, MI or cancer, hypertension, hypercholesterolemia, intake of total energy, alcohol, red or processed meat, fruit, and vegetables; and, in women, menopausal status and hormone use

Age, smoking, BMI, nonoccupational physical activity, history of hypertension, highest level of education, and energy intake

Age, sex, smoking, and social class

Sex, smoking, BMI, exercise, bread, fish, beef, fruit, and sweet desserts

Covariates in fully adjusted model

6 of 14 LUO ET AL

1 ADA, American Diabetes Association; AF, atrial fibrillation; AHS, The Adventist Health Study; BP, blood pressure; CVD, cardiovascular disease; FFQ, food-frequency questionnaire; FPG, fasting plasma glucose; HPFS, Health Professionals Follow-Up Study; IHD, ischemic heart disease; IWHS, Iowa Women’s Health Study; MI, myocardial infarction; NHS, Nurses’ Health Study; OGTT, oral-glucose-tolerance test; PHS, Physicians’ Health Study; ref, reference; PREDIMED, PREvencio´n con DIeta MEDiterra´nea; WHI-OS, Women’s Health Initiative Observational Study. 2 Exposure refers to how the data were collected.

Age, sex, and intervention group, BMI, smoke, educational, activity, history of diabetes, hypercholesterolemia, use of oral antidiabetic medication, antihypertensive medication, statins, and total energy intake, dietary variables, alcohol intake, and Mediterranean diet adherence Semiquantitative FFQ 137 items/by the end-point adjudication committee 55–80 Guasch-Ferre´ (22), 2013, the PREDIMED trial, Spain

Both

4.8

CVD death 81 all-cause mortality 323/31,077

0/wk; 1.00 (1.00, 1.00) ,1/wk; 0.84 (0.78, 0.90) 1/wk; 0.83 (0.76, 0.89) 2–4/wk; 0.79 (0.73, 0.86) $5/wk; 0.75 (0.62, 0.84) CVD death 0/wk; 1.00 (1.00, 1.00) 1–3/wk; 0.42 (0.24, 0.74) .3/wk; 0.45 (0.25, 0.81) All-cause mortality 0/wk; 1.00 (1.00, 1.00) 1–3/wk; 0.71 (0.54, 0.93) .3/wk; 0.61 (0.45, 0.83)

Covariates in fully adjusted model Age Sex First author (ref), year of publication, study name, location

TABLE 1 (Continued )

Follow-up time

No. of cases/subjects

Dietary category and RR (95% CI)

Exposure and case ascertainment2

META-ANALYSIS: NUTS AND DIABETES, CVD, AND MORTALITY

7 of 14

possible sources of heterogeneity [ie, sex, type of nut, IHD subtypes (fatal IHD or nonfatal MI), and stroke subtypes (ischemic and hemorrhagic)]. Sensitivity analyses were used to evaluate the effect of removing a single study from the analysis. Results of subgroup analyses are RRs (95% CIs) for comparison of highest with lowest categories of nut consumption because of a lack of data for person-years in some studies. BMI might mediate the relation between nut consumption and respective health outcomes. Whenever possible, we separately performed a meta-analysis on the multivariable-adjusted model with and without adjustment for BMI to explore the possible mediating effect of BMI on the relation of nut intake with health outcomes. In addition, we used restricted cubic splines with 4 knots at percentiles 5%, 35%, 65%, and 95% to evaluate potential linear curve associations between nut consumption and risk of diabetes, CVD, and all-cause mortality. Two-sided analyses were conducted with Stata 12.0 software (StataCorp). P , 0.05 was considered statistically significant.

RESULTS

Study characteristics We used a total of 31 reports from 18 studies in our main analysis; 14 studies were conducted in the United States, 1 study was conducted in China, 1 study was conducted in the Netherlands, 1 study was conducted in the United Kingdom, and 1 study was conducted in Spain. We identified 12,655 cases of type 2 diabetes, 8862 cases of CVD, 6623 cases of IHD, 6487 cases of stroke, and 48,818 cases of mortality. Characteristics of these studies are presented in Table 1 (13–30). Follow-up periods ranged from 4 to 30 y, and the age range was from 20 to 87 y. Nut consumption was assessed by food frequency questionnaires in all studies. Highest categories of nut intake ranged from $2 to $7 servings/wk; the lowest category of nut intake ranged from 0 to #1 serving/wk. Most of the studies included other dietary variables, such as protein sources and total caloric intake. All but 2 reports from one study (20) used multivariate models without adjustment for dietary factors, which could have caused confounding. Two studies (19, 21) examined the relation between intakes of peanut butter as well as all nuts and disease outcomes. Other studies examined peanuts and/or nuts without peanut butter. Seven studies (13, 18, 21–23, 25, 28) reported the relation between nut intake and other dietary and lifestyle factors at baseline. Participants who consumed more nuts were leaner, exercised more frequently, smoked less, and ate more fruit and vegetables, fish, whole grains, and dairy. Study quality scores ranged from 3 to 5. The mean quality score was 4.6. Nut consumption and diabetes Four reports reported age-adjusted estimates. Pooled results from these 4 reports suggested an inverse association between nut intake and diabetes (RR: 0.88; 95% CI: 0.84, 0.92; I2 = 0.0%, P = 0.60) for a 1-serving/d increment in nut consumption (see Supplemental Figure 1 under “Supplemental data” in the online issue). Two reports (one study with 2 large cohorts) reported multivariable-adjusted estimates without BMI. When we repeated the meta-analysis and used multivariable-adjusted estimates

8 of 14

LUO ET AL

FIGURE 2. Pooled RRs and 95% CIs for type 2 diabetes. We obtained pooled estimates by using a random-effects model. Dots indicate adjusted RRs by comparing high with low quantiles or per 7-serving/wk increase in nut intakes; the dashed line indicates no significant association between exposure and outcome; the diamonds indicate the pooled RRs; the size of the shaded square is proportional to the percentage weight of each study; horizontal lines indicate 95% CIs. NHS, Nurses’ Health Study; ref, reference.

without BMI, results were not substantially changed (RR: 0.80; 95% CI: 0.69, 0.94; I2 = 51.4%, P = 0.15) (Figure 2). Four reports reported multivariable-adjusted estimates with adjustment for BMI. When the primary meta-analysis was repeated by using fully adjusted estimates, the summary estimate was attenuated to null) RR: 1.03; 95% CI: 0.91, 1.16; I2 = 63.9%, P = 0.04; NS) (Figure 2). Pooled RRs for the comparison of extreme quantiles of nut intake for type 2 diabetes are also shown in Figure 2. Subgroup analyses were performed by sex, BMI, and type of nuts. From stratified analyses by type of nuts, we showed an inverse association between nut intake and diabetes in the multivariate model without BMI. In the BMI-adjusted model, the only remaining inverse association was with walnuts. When studies were stratified by BMI, studies conducted in populations with lower BMI (,25) suggested a borderline inverse association (RR: 0.68; 95% CI: 0.45, 1.01; I2 = 11.9%, P = 0.29). In studies conducted in populations with higher BMI ($25), the pooled RR was 0.78 (95% CI: 0.64, 0.95; I2 = 0%, P = 0.66) (Table 2). In the cubic spline model, we showed a nonlinear association between nut consumption and risk of type 2 diabetes by using data with multivariable-adjustment without adjustment for BMI (Figure 3A; P-nonlinearity , 0.001). However, with the use of data with full adjustment, we showed no significant association between nut consumption and risk of diabetes (Figure 3B; P-nonlinearity = 0.746).

Nut intake, IHD, and CVD Results from the random-effects meta-analysis of the relation between nut intake and incidences of IHD and CVD are shown in Figure 4. For a 1-serving/d increment in nut consumption, pooled RRs (95% CIs) were 0.72 (0.64, 0.81) for IHD with no between-study heterogeneity (I2 = 0%, P = 0.64) and 0.71 (0.59, 0.85) for overall CVD with modest between-study heterogeneity (I2 = 48.8%, P = 0.12) (Figure 4). After the removal of estimates that included peanut butter and all nuts, pooled RR estimates were 0.73 (95% CI: 0.64, 0.83) for IHD and 0.64 (95% CI: 0.43, 0.97) for CVD, respectively. With the exclusion of estimates without adjustment for dietary factors, the pooled RR for IHD was 0.73 (95% CI: 0.64, 0.83). Pooled RRs for the comparison of extreme quantiles of nut intake were 0.66 (95% CI: 0.55, 0.78; I2 = 62.5%, P = 0.02) for IHD and 0.70 (95% CI: 0.60, 0.81; I2 = 22.8%, P = 0.27) for CVD (Figure 4). In stratified analyses by sex, an inverse relation between nut consumption and IHD was shown in women (RR: 0.68; 95% CI: 0.61, 0.76), with the I2 value decreasing from 62.5% to 0.0%, but not in men (RR: 0.87; 95% CI: 0.59, 1.28; I2 = 71.5%, P = 0.06; NS). For the type of IHD, nut intake was inversely associated with fatal IHD (RR: 0.68; 95% CI: 0.62, 0.75; I2 = 0.0%, P = 0.54) but not nonfatal MI (RR: 0.70; 95% CI: 0.43, 1.17; I2 = 75.7%, P = 0.02; NS). Similar results were shown in studies with and without peanut butter (Table 2). In

9 of 14

META-ANALYSIS: NUTS AND DIABETES, CVD, AND MORTALITY TABLE 2 Subgroup analyses (highest compared with lowest categories)1 Homogeneity Source Subgroup analyses for type 2 diabetes Multivariate-adjusted model without BMI Type of nuts Total tree nuts Walnuts Multivariate-adjusted model + BMI Type of nuts Total tree nuts Walnuts Sex F M BMI ,25 kg/m2 $25 kg/m2 Subgroup analyses for IHD2 Multivariate-adjusted model + BMI IHD subgroup Fatal IHD Nonfatal MI Sex F M Type of nuts Peanut butter + nuts Nuts Subgroup analyses for stroke Multivariate-adjusted model + BMI Stroke subgroup Ischemic stroke Hemorrhagic stroke Sex F M Subgroup analyses for all-cause mortality Multivariate-adjusted model + BMI Sex F M Type of nuts Peanut butter + nuts Nuts 1

No. of reports

RR (95% CI)

P-heterogeneity

I2 (%)

2 2

0.85 (0.75, 0.95) 0.67 (0.54, 0.82)

— —

— —

2 2

0.98 (0.87, 1.10) 0.76 (0.62, 0.94)

— —

— —

3 1

0.94 (0.77, 1.15) 0.87 (0.61, 1.24)

0.03 —

70.8 —

2 2

0.68 (0.45, 1.01) 0.78 (0.64, 0.95)

0.29 0.66

11.9 0

5 3

0.68 (0.62, 0.75) 0.70 (0.43, 1.17)

0.54 0.02

0 75.7

3 2

0.68 (0.61, 0.76) 0.87 (0.59, 1.28)

0.74 0.06

0 71.5

2 4

0.66 (0.52, 0.84) 0.70 (0.55, 0.89)

,0.01 0.59

75.7 0

3 2

0.95 (0.84, 1.08) 1.17 (0.54, 2.54)

0.87 0.04

0 77.3

2 2

0.87 (0.77, 0.98) 0.95 (0.82, 1.11)

0.84 0.40

0 0

5 4

0.88 (0.81, 0.96) 0.85 (0.75, 0.95)

0.07 0.05

54.2 60.7

1 10

0.89 (0.81, 0.99) 0.85 (0.79, 0.91)

— ,0.01

— 60.2

Pooled estimates were obtained by using a random-effects model. IHD, ischemic heart disease.

the cubic spline model, we showed a nonlinear association between nut consumption and risk of CVD (Figure 3C; P-nonlinearity , 0.001). Nut intake and stroke The pooled RR of total stroke for the comparison of extreme quantiles of nut intake was 0.91 (95% CI: 0.81, 1.02; I2 = 20.4%, P = 0.285; NS) (Figure 4). In a stratified analysis, an inverse association between nut consumption and stroke was shown in women (RR: 0.87; 95% CI: 0.77, 0.98; I2 = 0%, P = 0.84) but not men (RR: 0.95; 95% CI: 0.82, 1.11; I2 = 0%, P = 0.40; NS). In addition, pooled RRs were not statistically significant for either stroke type (Table 2).

Nut intake and all-cause mortality For all-cause mortality, the pooled RR with a 1-serving/d increment in nut intake was 0.83 (95% CI: 0.76, 0.91) with moderately high heterogeneity (I2 = 62.1%, P = 0.032) (Figure 5). After the exclusion of estimates that included peanut butter as well as all nuts, the pooled RR was 0.80 (95% CI: 0.71, 0.90). When we excluded estimates without adjustment for dietary factors, the pooled RR was 0.84 (95% CI: 0.76, 0.92). The pooled RR for the comparison of extreme quantiles of nut intake was 0.85 (95% CI: 0.79, 0.91) for all-cause mortality with moderately high heterogeneity (I2 = 60.2%, P = 0.005) (Figure 5). Similar results were shown in studies conducted in women and men and in studies with and without peanut butter (Table 2).

10 of 14

LUO ET AL

FIGURE 3. Dose-response relation plots between nut intakes (servings/wk) and risk of type 2 diabetes (A and B), cardiovascular disease (C), and all-cause mortality (D) were estimated by using random-effects metaregression. Dotted lines represent the 95% CIs for the fitted trend. RR estimates with multivariableadjustment except adjustment for BMI were used in panel A; RR estimates with multivariable-adjustment were used in panels B–D. CVD, cardiovascular disease.

In the cubic spline model, we showed a nonlinear association between nut consumption and risk of all-cause mortality (Figure 3D; P-nonlinearity = ,0.001). Sensitivity analysis When we recalculated the overall homogeneity and effect size by removing one report at a time, we showed that none of the single studies substantially changed the pooled estimates for IHD, overall CVD, and mortality (see Supplemental Table 1 under “Supplemental data” in the online issue). For stroke, the pooled estimate changed from 0.91 (95% CI: 0.81, 1.02) to 0.89 (95% CI: 0.80, 0.98; I2 = 0.0%, P = 0.932) when the hemorrhagic stroke result from the Physicians’ Health Study (23) was excluded. For diabetes, the pooled estimate was not substantially influenced with the omission of the study of Parker et al (26), which was from correspondence and not a peer-reviewed publication. The omission of the other studies one at a time did not materially change the pooled estimate. Assessment of publication bias No significant evidence of substantial publication bias was observed [P = 1.00 for diabetes, P = 1.00 for IHD, P = 0.31 for

CVD, P = 0.46 for mortality (high compared with low analyses) and P = 0.81 for diabetes, P = 1.00 for IHD, P = 0.09 for CVD, P = 0.31 for stroke, and P = 0.21 for mortality (dose-response analyses)]. The assessment of publication bias was based on the fully adjusted model.

DISCUSSION

This meta-analysis indicated that nut intake is inversely associated with IHD, overall CVD, stroke in women, and all-cause mortality but not with diabetes and total stroke in the fully adjusted model. A significant inverse association seen between nut intake and type 2 diabetes was attenuated by adjustment for BMI, which indicated that the association was largely mediated through BMI. Frequent nut intake may improve health outcomes via multiple mechanisms. First, nuts are high in fat, but most of the fat is unsaturated fatty acids (39) that can reduce risk of CVD and diabetes (40, 41). Second, nuts contain few carbohydrates and, thus, contribute little to postprandial glycemia (42). Third, inflammation has been linked to risk of CVD and diabetes (43), and evidence has suggested that frequent nut consumption has an inverse association with circulating inflammatory cytokines and a positive relation with plasma adiponectin (44). Fourth, nuts have high amounts of protein, L-arginine, folate, fiber, and

META-ANALYSIS: NUTS AND DIABETES, CVD, AND MORTALITY

11 of 14

FIGURE 4. Pooled RRs and 95% CIs for IHD, stroke, and CVD. We obtained pooled estimates by using a random-effects model. Dots indicate adjusted RRs by comparing high with low quantiles or per 7-serving/wk increase in nut intakes; the dashed line indicates no significant association between exposure and outcome; the diamonds indicate the pooled RRs; the size of the shaded square is proportional to the percentage weight of each study; and horizontal lines indicate 95% CIs. CVD, cardiovascular disease; IHD, ischemic heart disease; ref, reference; W, women.

phytosterols (39). Unsalted varieties of nuts have low sodium contents. Nuts are also rich in minerals (eg, calcium, magnesium, and potassium) (39), which are associated with decreased overall cardiovascular risk (45). In addition, frequent nut intake can improve lipid profiles. A recent pooled analysis of 25 clinical studies on various kinds of nuts showed a dose-response cholesterol-lowering effect. With an average daily intake of 67 g nuts, mean reductions in total and LDL cholesterol were 11 mg/dL and 10 mg/dL, respectively (8). Another meta-analysis of 13 clinical trials indicated that diets supplemented with walnuts were associated with a 13.7% decrease in LDL cholesterol (46). Because a large body of consistent evidence has supported the beneficial effects of frequent nut intake on various intermediate mediators of chronic diseases, it is plausible that nuts protect against IHD, overall CVD, and all-cause mortality. Heterogeneity was present in our analyses. The heterogeneity for IHD and stroke were partly explained, and some meaningful

results were obtained in subgroup analyses. For both IHD and stroke, results were stronger in women than men. These observed differences may be the result of a different hormonal milieu between sexes or just due to chance because of limited study. By IHD type, the result was stronger for fatal IHD than for MI. Data from Albert et al (25) showed a significant inverse relation between nut intake and sudden cardiac death but no significant inverse relation between nut intake and nonsudden IHD death or nonfatal MI. Separate analyses by stroke type did not show significant effects for either ischemic or hemorrhagic stroke. Greater intake of nuts has been associated with improved BP in a previous meta-analysis (47). If nuts decrease BP, it seems logical that frequent nut consumption would reduce the incidence of ischemic stroke. Our nonsignificant finding on ischemic stroke may have been attributable to the small number of cases in included studies, which limited the power of the analysis. In addition, stroke is a heterogeneous disease that is difficult to accurately

12 of 14

LUO ET AL

FIGURE 5. Pooled RRs and 95% CIs for all-cause mortality. We obtained pooled estimates by using a random-effects model. Dots indicate adjusted RRs by comparing high with low quantiles or per a 7-serving/wk increase in nut intakes; the dashed line indicates no significant association between exposure and outcome; the diamonds indicate the pooled RRs; the size of the shaded square is proportional to the percentage weight of each study; and horizontal lines indicate 95% CIs. ref, reference; W, women.

diagnose. Additional studies are warranted to investigate the association between the consumption of nuts and ischemic stroke and investigate the inverse relation between nut intake and stroke in women. Long-term, double-blind, randomized controlled trials provide the best evidence on the effect between nutrients and disease. In the PREvencio´n con DIeta MEDiterra´nea (PREDIMED) trial, Mediterranean diets supplemented with 30 g nuts/d (almonds, hazelnuts, and walnuts) significantly reduced systolic BP (48), risk of metabolic syndrome (49), and incidence of diabetes (50) compared with a low-fat diet. Recently, long-term results from this trial showed reduced risk of stroke, total CVD, and all-cause mortality for the group assigned to the Mediterranean diet with mixed nuts compared with the control group (51). A recent meta-analysis showed no association between nut consumption and diabetes (47). However, the results did not consider models with and without adjustment for BMI. Obesity is the most-important determinant of type 2 diabetes. Previous studies have shown that frequent nut consumption is associated with less weight gain (52). Thus, BMI can be considered a mediating variable in the association between nut intake and diabetes. The inverse association between walnut consumption and diabetes risk remained significant even after adjustment for BMI. Note that walnuts have the highest a-linolenic acid content of all nuts (53). In addition, Guasch-Ferre´ et al (22) showed an inverse relation between walnuts and cancer mortality but not with other nuts in the PREvencio´n con DIeta MEDiterra´nea trial and concluded this effect could be attributable to the rich free and total polyphenols contents in walnuts compared with other nuts. More studies are needed to elucidate the effect of different types

of nuts separately on multiple health outcomes. Note the inverse associations between frequent nut consumption and risk of CVD and total mortality were not altered substantially after adjustment for BMI. Strengths and limitations We included prospective cohort studies with high-quality scores, large sample sizes, and long-term follow-ups. We conducted several sensitivity analyses to evaluate the robustness of our findings. In addition, we obtained as much data as possible for dose-response analyses, which complemented the results from categorical analyses. Our meta-analysis had several limitations. First, we may have overlooked some studies and did not include unpublished or non– English-language reports. Second, there are many kinds of nuts, with varying effects on health outcomes. However, we could only single out walnuts and peanuts. Our findings should be interpreted as the average associations with different types of nuts. Third, various preparation and processing methods may alter the availability of bioactive compounds in nuts (54), but it is beyond the scope of large epidemiologic studies to assess these factors in modifying health effects of nuts. Fourth, significant heterogeneity was present for the meta-analyses of nut consumption and mortality. Sources of heterogeneity were not completely clear, but they might be due to different types and amounts of nuts consumed in different populations. Finally, observational studies cannot exclude the effects of some unknown confounding factors or residual confounding attributable to other dietary and lifestyle factors because nut eaters tend to

META-ANALYSIS: NUTS AND DIABETES, CVD, AND MORTALITY

follow a healthy dietary pattern. Therefore, it is difficult to tease out the independent effects of nut consumption from other dietary factors in observational analyses. For this reason, our results should be interpreted in the context of mechanistic studies and human intervention studies. In conclusion, our meta-analysis indicates significant inverse associations between frequent nut intake and IHD, overall CVD, and all-cause mortality. These findings support recommendations to include nuts as part of healthy dietary patterns for the prevention of chronic diseases. We thank Adam M Bernstein for providing data for the meta-analysis. The authors’ responsibilities were as follows—LL and FBH: were the guarantors, had full access to all study data, took responsibility for the study integrity and accuracy of data analyses, designed the study, and reviewed or revised the final manuscript; CL: led the conception and drafting of the protocol, analyzed data, drafted the manuscript, provided a critical review of the content of the manuscript, and revised the manuscript for important intellectual content; YZ and YD: searched databases according to inclusion and exclusion criteria, reviewed search results by titles and abstracts, and retrieved full-text articles to identify eligible trials; ZS and SC: extracted data and evaluated the methodological quality by using criteria that were previously established; MY: settled discrepancies by discussion in accordance with the authors’ selection criteria and gave advice on the metaanalysis methodology; and all authors: contributed to the manuscript. None of the authors had a conflict of interest.

REFERENCES 1. Foot DK, Lewis RP, Pearson TA, Beller GA. Demographics and cardiology, 1950-2050. J Am Coll Cardiol 2000;35:1067–81. 2. Taylor TN, Davis PH, Torner JC, Holmes J, Meyer JW, Jacobson MF. Lifetime cost of stroke in the United States. Stroke 1996;27:1459–66. 3. Diringer MN, Edwards DF, Mattson DT, Akins PT, Sheedy CW, Hsu CY, Dromerick AW. Predictors of acute hospital costs for treatment of ischemic stroke in an academic center. Stroke 1999;30:724–8. 4. Shaw JE, Sicree RA, Zimmet PZ. Global estimates of the prevalence of diabetes for 2010 and 2030. Diabetes Res Clin Pract 2010;87:4–14. 5. Sabate´ J, Ros E, Salas-Salvado´ J. Nuts: nutrition and health outcomes. Preface. Br J Nutr 2006;96(suppl 2):S1–2. 6. Brufau G, Boatella J, Rafecas M. Nuts: source of energy and macronutrients. Br J Nutr 2006;96(suppl 2):S24–8. 7. Dreher ML, Maher CV, Kearney P. The traditional and emerging role of nuts in healthful diets. Nutr Rev 1996;54:241–5. 8. Sabate´ J, Oda K, Ros E. Nut consumption and blood lipid levels: a pooled analysis of 25 intervention trials. Arch Intern Med 2010;170: 821–7. 9. Jiang R, Jacobs DR Jr, Mayer-Davis E, Szklo M, Herrington D, Jenny NS, Kronmal R, Barr RG. Nut and seed consumption and inflammatory markers in the multi-ethnic study of atherosclerosis. Am J Epidemiol 2006;163:222–31. 10. Wien MA, Sabate´ JM, Ikle DN, Cole SE, Kandeel FR. Almonds vs complex carbohydrates in a weight reduction program. Int J Obes Relat Metab Disord 2003;27:1365–72. 11. Casas-Agustench P, Lopez-Uriarte P, Ros E, Bullo M, Salas-Salvado´ J. Nuts, hypertension and endothelial function. Nutr Metab Cardiovasc Dis 2011;21(suppl 1):S21–33. 12. Kris-Etherton PM, Hu FB, Ros E, Sabate´ J. The role of tree nuts and peanuts in the prevention of coronary heart disease: multiple potential mechanisms. J Nutr 2008;138:1746S–51S. 13. Bao Y, Han J, Hu FB, Giovannucci EL, Stampfer MJ, Willett WC, Fuchs CS. Association of nut consumption with total and cause-specific mortality. N Engl J Med 2013;369:2001–11. 14. Fraser GE, Sumbureru D, Pribis P, Neil RL, Frankson MA. Association among health habits, risk factors, and all-cause mortality in a black California population. Epidemiology 1997;8:168–74. 15. Bernstein AM, Pan A, Rexrode KM, Stampfer M, Hu FB, Mozaffarian D, Willett WC. Dietary protein sources and the risk of stroke in men and women. Stroke 2012;43:637–44.

13 of 14

16. Fraser GE, Shavlik DJ. Risk factors for all-cause and coronary heart disease mortality in the oldest-old. The Adventist Health Study. Arch Intern Med 1997;157:2249–58. 17. van den Brandt PA. The impact of a Mediterranean diet and healthy lifestyle on premature mortality in men and women. Am J Clin Nutr 2011;94:913–20. 18. Pan A, Sun Q, Manson JE, Willett WC, Hu FB. Walnut consumption is associated with lower risk of type 2 diabetes in women. J Nutr 2013; 143:512–8. 19. Blomhoff R, Carlsen MH, Andersen LF, Jacobs DR Jr. Health benefits of nuts: potential role of antioxidants. Br J Nutr 2006;96(suppl 2):S52–60. 20. Mann JI, Appleby PN, Key TJ, Thorogood M. Dietary determinants of ischaemic heart disease in health conscious individuals. Heart 1997;78: 450–5. 21. Li TY, Brennan AM, Wedick NM, Mantzoros C, Rifai N, Hu FB. Regular consumption of nuts is associated with a lower risk of cardiovascular disease in women with type 2 diabetes. J Nutr 2009;139:1333–8. 22. Guasch-Ferre´ M, Bullo M, Martinez-Gonzalez MA, Ros E, Corella D, Estruch R, Fito M, Aros F, Warnberg J, Fiol M, et al. Frequency of nut consumption and mortality risk in the PREDIMED nutrition intervention trial. BMC Med 2013;11:164. 23. Djousse´ L, Gaziano JM, Kase CS, Kurth T. Nut consumption and risk of stroke in US male physicians. Clin Nutr 2010;29:605–9. 24. Fraser GE, Sabate´ J, Beeson WL, Strahan TM. A possible protective effect of nut consumption on risk of coronary heart disease. The Adventist Health Study. Arch Intern Med 1992;152:1416–24. 25. Albert CM, Gaziano JM, Willett WC, Manson JE. Nut consumption and decreased risk of sudden cardiac death in the Physicians’ Health Study. Arch Intern Med 2002;162:1382–7. 26. Parker ED, Harnack LJ, Folsom AR. Nut consumption and risk of type 2 diabetes. JAMA 2003;290:38–9; author reply 9–40. 27. Villegas R, Gao YT, Yang G, Li HL, Elasy TA, Zheng W, Shu XO. Legume and soy food intake and the incidence of type 2 diabetes in the Shanghai Women’s Health Study. Am J Clin Nutr 2008;87:162–7. 28. Kochar J, Gaziano JM, Djousse´ L. Nut consumption and risk of type II diabetes in the Physicians’ Health Study. Eur J Clin Nutr 2010;64:75–9. 29. Bernstein AM, Sun Q, Hu FB, Stampfer MJ, Manson JE, Willett WC. Major dietary protein sources and risk of coronary heart disease in women. Circulation 2010;122:876–83. 30. Yaemsiri S, Sen S, Tinker L, Rosamond W, Wassertheil-Smoller S, He K. Trans fat, aspirin, and ischemic stroke in postmenopausal women. Ann Neurol 2012;72:704–15. 31. Jiang R, Manson JE, Stampfer MJ, Liu S, Willett WC, Hu FB. Nut and peanut butter consumption and risk of type 2 diabetes in women. JAMA 2002;288:2554–60. 32. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, Moher D, Becker BJ, Sipe TA, Thacker SB. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Metaanalysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000;283:2008–12. 33. Greenland S, Longnecker MP. Methods for trend estimation from summarized dose-response data, with applications to meta-analysis. Am J Epidemiol 1992;135:1301–9. 34. Orsini N, Bellocco R. S G. Generalized least squares for tend estimation of summarized dose-response data. Stata J 2006;6:40–57. 35. Food portion sizes. 3rd ed. Norwich, London: TSO, 2005. 36. Higgins JP. Commentary: heterogeneity in meta-analysis should be expected and appropriately quantified. Int J Epidemiol 2008;37:1158–60. 37. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003;327:557–60. 38. Egger M, Davey Smith G, Schneider M, Minder C. Bias in metaanalysis detected by a simple, graphical test. BMJ 1997;315:629–34. 39. Ros E. Health benefits of nut consumption. Nutrients 2010;2:652–82. 40. Pan A, Chen M, Chowdhury R, Wu JH, Sun Q, Campos H, Mozaffarian D, Hu FB. alpha-Linolenic acid and risk of cardiovascular disease: a systematic review and meta-analysis. Am J Clin Nutr 2012;96:1262–73. 41. Wu JH, Micha R, Imamura F, Pan A, Biggs ML, Ajaz O, Djousse´ L, Hu FB, Mozaffarian D. Omega-3 fatty acids and incident type 2 diabetes: a systematic review and meta-analysis. Br J Nutr 2012;107(suppl 2): S214–27. 42. Josse AR, Kendall CW, Augustin LS, Ellis PR, Jenkins DJ. Almonds and postprandial glycemia–a dose-response study. Metabolism 2007; 56:400–4.

14 of 14

LUO ET AL

43. Pradhan AD, Ridker PM. Do atherosclerosis and type 2 diabetes share a common inflammatory basis? Eur Heart J 2002;23:831–4. 44. Ros E. Nuts and novel biomarkers of cardiovascular disease. Am J Clin Nutr 2009;89:1649S–56S. 45. Cordain L, Eaton SB, Sebastian A, Mann N, Lindeberg S, Watkins BA, O’Keefe JH, Brand-Miller J. Origins and evolution of the Western diet: health implications for the 21st century. Am J Clin Nutr 2005;81:341–54. 46. Banel DK, Hu FB. Effects of walnut consumption on blood lipids and other cardiovascular risk factors: a meta-analysis and systematic review. Am J Clin Nutr 2009;90:56–63. 47. Guo K, Jiang Y, Zhou Z, Li Y. RETRACTED: Nut consumption with risk of hypertension and type 2 diabetes mellitus: a meta-analysis of prospective cohort studies. Eur J Prev Cardiol, (Epub ahead of print 30 August 2013). 48. Estruch R, Martinez-Gonzalez MA, Corella D, Salas-Salvado´ J, RuizGutierrez V, Covas MI, Fiol M, Gomez-Gracia E, Lopez-Sabater MC, Vinyoles E, et al. Effects of a Mediterranean-style diet on cardiovascular risk factors: a randomized trial. Ann Intern Med 2006;145:1–11. 49. Salas-Salvado´ J, Fernandez-Ballart J, Ros E, Martinez-Gonzalez MA, Fito M, Estruch R, Corella D, Fiol M, Gomez-Gracia E, Aros F, et al. Effect of a Mediterranean diet supplemented with nuts on metabolic

50.

51.

52.

53. 54.

syndrome status: one-year results of the PREDIMED randomized trial. Arch Intern Med 2008;168:2449–58. Salas-Salvado´ J, Bullo M, Babio N, Martinez-Gonzalez MA, IbarrolaJurado N, Basora J, Estruch R, Covas MI, Corella D, Aros F, et al. Reduction in the incidence of type 2 diabetes with the Mediterranean diet: results of the PREDIMED-Reus nutrition intervention randomized trial. Diabetes Care 2011;34:14–9. Estruch R, Ros E, Salas-Salvado´ J, Covas MI, Corella D, Aros F, Gomez-Gracia E, Ruiz-Gutierrez V, Fiol M, Lapetra J, et al. Primary prevention of cardiovascular disease with a Mediterranean diet. N Engl J Med 2013;368:1279–90. Bes-Rastrollo M, Wedick NM, Martinez-Gonzalez MA, Li TY, Sampson L, Hu FB. Prospective study of nut consumption, long-term weight change, and obesity risk in women. Am J Clin Nutr 2009;89: 1913–9. Ros E, Hu FB. Consumption of plant seeds and cardiovascular health: epidemiological and clinical trial evidence. Circulation 2013;128:553– 65. Kendall CW, Josse AR, Esfahani A, Jenkins DJ. Nuts, metabolic syndrome and diabetes. Br J Nutr 2010;104:465–73.

Nut consumption and risk of type 2 diabetes, cardiovascular disease, and all-cause mortality: a systematic review and meta-analysis.

Epidemiologic studies have shown inverse associations between nut consumption and diabetes, cardiovascular disease (CVD), and all-cause mortality, but...
1MB Sizes 2 Downloads 3 Views